Dear Stata List-
I am conducting a secondary data analysis of survey responses and am struggling with a two issues:
Background:
The set is 3 surveys that have been merged. The first survey has the dependent variable and was sent out monthly, the second ones yearly over 2 years contain the independent variables (stable demographic variables).
There is highly variable non-response. Subjects may have answered one survey and not the others- or not answered all the questions in the surveys they responded to. Also the survey distribution was based on subjects participating in a certain activity. Some participated only once and were given one monthly survey--others participated several times and received several surveys.
The questions:
1. For the subjects that responded multiple times to the dependent variable question there is some variability in their responses. I would like to capture this variability- but only count each subject once. I have been reading some about weights--but am not sure which is the correct weight to use and if this is the best approach. I wonder if it would be better to analyze as panel data, or if the survey commands and panel commands can be used together.
2. I would like to include in the analysis any subject who answered the monthly survey at least once. To be able to examine for possible response bias I need to be able to separate non-responders from responders and make comparisons--but in this case there is overlap. Do you know an efficient way to separate the "never responders" from the group that responded at least once?
I have access to Stata 8 & 9. I have attached an excerpt of the data for you reference.
I would greatly appreciate any insights you have for me.
id year svryyr mn returned dependent var
201 2003 2002 7 Yes:1 Somewhat Satisfied
201 2003 2003 1 Yes:1 Neither Satisfied nor Dissatisfied
201 2002 2001 10 Yes:1 Very Satisfied
205 2003 2003 5 No:0 .q
205 2002 2002 5 No:0 .q
205 2003 2002 10 Yes:1 Somewhat Dissatisfied
205 2002 2002 1 Yes:1 Somewhat Satisfied
207 2002 2001 9 Yes:1 Somewhat Satisfied
207 2003 2002 7 No:0 .q
208 2003 2002 9 Yes:1 Neither Satisfied nor Dissatisfied
208 2002 2002 1 Yes:1 Somewhat Dissatisfied
208 2002 2001 7 No:0 .q
213 2002 2001 11 No:0 .q
213 2003 2003 4 No:0 .q
214 2003 2002 8 Yes:1 Somewhat Satisfied
214 2002 2001 8 Yes:1 Neither Satisfied nor Dissatisfied
215 2002 2001 7 No:0 .q
221 2002 2002 6 Yes:1 Somewhat Satisfied
223 2003 2002 7 Yes:1 Somewhat Satisfied
223 2002 2002 4 Yes:1 Somewhat Satisfied
225 2002 2001 8 No:0 .q
225 2003 2003 4 No:0 .q
225 2002 2001 12 Yes:1 Neither Satisfied nor Dissatisfied
228 2002 2002 3 Yes:1 Somewhat Satisfied
228 2003 2003 3 Yes:1 Neither Satisfied nor Dissatisfied
231 2002 2001 8 Yes:1 Very Satisfied
234 2002 2001 7 No:0 .q
234 2003 2003 5 Yes:1 Somewhat Satisfied
235 2003 2003 4 No:0 .q
235 2002 2002 5 Yes:1 Somewhat Satisfied
236 2003 2003 4 No:0 .q